The use of uniform structures in adaptive dynamic control of robots

In the traditional approaches of computed torque control and its robust/adaptive versions development of complicated dynamic models tailored to the particular systems under consideration and based on the Denavit-Hartenberg conventions is prevailing. In the practice, due to the presence of unmodeled external dynamic interactions, in spite of their high complexity, these dynamic models are only of approximate nature. The aim of this paper is to give a report on a novel approach replacing the analytical models with uniform structures of tunable parameters and “standardized” algorithms not depending on the particular properties of the system under control. Unlike the traditional soft computing approaches gaining these structures from artificial neural network architectures or the membership functions of fuzzy controllers, the present work uses the more simple and definite “inventory” of classical mechanics and group theory for this purpose. Simulation results gained for a 3 DOF SCARA arm are presented for illustrating the efficiency of the method